Aspen

Members:

Overview and Goals:

Aspen is a research project dealing with sensor integration which aims to find a way to use sensors connected to a network to summarize data and produce definite results or solutions.  Based on the data received from different sensors, Aspen will be able to determine the status of the world, and to help people solve problems.

Examples of Situations where Aspen can be used:

Imagine the emergency room at South Lake Hospital.  A horrible tropical storm rages outside and accident victims swarm the ER.  A woman is fighting for her life, but wait-- the ER cannot find the ventilator or Dr. Ginsburg, the attending surgeon on duty.  Nurse Johnson at the nursing station pages Dr. Ginsburg but he is not responding to his pages.  Nurse Johnson very calmly turns to her computer and types in "Dr. Ginsburg" and the name of the ventilator.  Magically dots appear on the screen revealing the locations of both man and equipment on a hospital floor map.  The ventilator is stashed in the corner of ER #3.  Dr. Ginsburg is in the restroom where there are no speakers; his pager was knocked out by the storm.  RFID sensors located on Dr. Ginsburg’s ID card and on the ventilator provided coordinates for their position, which was then summarized, and displayed digitally on the monitor.  The nurse dispatches two orderlies to the men’s room and ER #3 to retrieve Dr. Ginsburg and the ventilator.     

Meanwhile in another wing of the hospital, a woman approaches the information desk to find out the condition of her best friend, Jane Allen.  Jane had been in an accident five hours earlier and had been brought to South Lake.  The woman at the desk typed Jane Allen into the computer and discovers her condition and location.  The computer reveals that Jane’s heart monitor had flat lined, and that she was receiving emergency treatment to save her life.  The information desk attendant was able to receive advance notice about Jane, via the data reported from the sensors connected to the machines in Jane’s room monitoring her condition, and send her friend to the waiting room until further notice.  This allows for the hospital to keep the patient’s friends and family up to date on her condition.

Across the street, a shipment of Britney Spears’ new CD has just arrived at the local Best Buy.   In a few short hours the store will be overflowing with little girls anxious to hear Spears’ new hit song. Ned, the manager, needs to get all of the boxes of CDs logged for inventory and placed on the shelves.  He wheels the crate under the giant sensor on the ceiling, and clicks “read” on his computer.  A spreadsheet appears accounting for 500 Britney Spears CDs.  But wait--the manager had ordered 750 copies not 500.  Because this process was so quick and simple, Ned was able to catch the delivery truck before it pulled off to get his other crate of CDs.  Suddenly everything is better, and another 250 little girls will have the delightful experience of Britney Spears.

Across town, scientists in a lab research Florida’s American Alligator population.  They want to monitor them in their natural environment, but realize that the Alligators will behave very hostilely if a bunch of scientists physically follow their every move.  What if the scientists could implant electronic monitoring devices on the alligators?  They could monitor their relative positions via a Global Positioning System (GPS), moisture sensors could determine whether or not they are in the water, and cameras would reveal pictures of their surroundings.  This data would be invaluable to scientists studying the natural behavior of wild American Alligators. 

The Aspen technology will allow people such as scientists, doctors, and hospital technicians to monitor behaviors and conditions similar to the situations above.  Aspen will stream together data collected from the various devices in each situation to allow for easier transfer across a wireless network.

Aspen Details:

Aspen can be broken down into two parts: security and sensor/stream integration.  The security part aims to establish ways to secure communication between wireless devices. Part of this research includes exploring how difficult it is to hack into networks and what measures are necessary to capture traffic within a network.  To enable streaming integration the research team is designing algorithms to efficiently join data from different sensors.  Sensor and stream integration tries to integrate heterogeneous data from a number of different types of wireless sensors.  Cameras, temperature sensors, audio devices, RFID readers and motion sensors collect data.  How can this data from independent devices be joined to detect high-level objects (e.g., passengers) or behaviors (e.g., walking)?  Sending data in a wireless network is expensive.  By aggregating within the network things become easier and more efficient.  By condensing the data, the least possible amount of data can be sent while still preserving the essence of the data; which is what is most important.

The Aspen project will help everyone from doctors to retail employees and research scientists.  As the examples show, the data from these ordinary devices can be manipulated and combined to improve everyday occurrences.

Publications:

Funding Resources:

Funded in part by a seed grant from ISTAR, the Penn Institute for Strategic Threat Analysis and Response.

Members:

Overview and Goals:

Traditional Data Integration combines integrated data from multiple independent and heterogeneous data sources.  In most cases the data is displayed in different formats making it very difficult for the computer to combine the information from the various sources.  Traditional data integration takes the data from different databases and sends it to a central server where it is combined and displayed in a new virtual database.  However, problems occur when the data does not agree, or if the source of the data is unreliable.  If there is an error in the data or the databases contain different quantities for the same category, the computer does not know who to trust.  These problems and more make data integration a very difficult task. Orchestra resolves some of these issues.

Orchestra allows data integration between multiple independent and heterogeneous data sources when there is an error in some of the data or if different sites simply disagree.  Instead of having a central computer combining all of the data and spitting it out to a virtual database, Orchestra sends the group’s data to the individual users according to "policies" specifying what data they trust.  When a change is made in the data, it is sent to everyone else who then can choose whether or not to accept the change based on the reliability of the source.  This example of peer to peer data exchange/ management creates a more individual/ specific way to display data.

Everyday applications of Orchestra:

Imagine these three everyday occurrences:

In situation 1, the address information is very similar, but contains slightly different data.  In situation 2, the databases contain the same information, except one prefers different abbreviation styles, and one uses endnotes while the other uses footnotes.  In situation 3, Geneticists Z, Y, and X only disagree about certain gene locations.  Orchestra realizes that a unified view of different sources is impossible since the data may disagree.  It instead records the updates that the individual sources make and sends them to all of the other sources.  Each source then can choose which updates to apply to itself via the individual user-specified policy.  This means that all data sharing operations involve only one peer; i.e. all data sharing is between individual sources.

Publications:

Funding Resources:

This research is funded by NSF CAREER grant award #IIS-0477972, awarded to Zachary G. Ives at the University of Pennsylvania, and NSF SEIII grant #IIS-0513778.